Text-to-vectorized representation transformation

Abstract

An approach for a fast and accurate word embedding model, “desc2vec,” for out-of-dictionary (OOD) words with a model learning from the dictionary descriptions of the word is disclosed. The approach includes determining that a target text element is not in a set of reference text elements, information describing the target text element is obtained. The information comprises a set of descriptive text elements. A set of vectorized representations for the set of descriptive text elements is determined. A target vectorized representation for the target text element is determined based on the set of vectorized representations using a machine learning model. The machine learning model is trained to represent a predetermined association between the set of vectorized representations for the set of descriptive text elements describing the target text element and the target vectorized representation.

What is claimed is: 
     
       1. A computer-implemented method comprising:
 in accordance with a determination that a target text element is not in a set of reference text elements, obtaining, by one or more processors, information describing the target text element, the information comprising a set of descriptive text elements and 
 wherein the set of reference text elements are mapped to a set of vectorized presentations based on a further predetermined association and wherein the further predetermined association is represented by a first machine learning model trained based on a corpus comprising the set of reference text elements; 
 determining, by the one or more processors, a set of vectorized representations for the set of descriptive text elements further comprises:
 determining, by the one or more processors, the set of vectorized representations for the set of descriptive text elements based on the further predetermined association, respectively; and 
 
 determining, by the one or more processors, a target vectorized representation for the target text element based on the set of vectorized representations using a second machine learning model, the second machine learning model being trained to represent a predetermined association between the set of vectorized representations for the set of descriptive text elements describing the target text element and the target vectorized representation. 
 
     
     
       2. The computer-implemented method according to  claim 1 , wherein obtaining the information describing the target text element comprises:
 in accordance with a determination that the target text element is not in the set of reference text elements, determining, by the one or more processors, a synonymous text element for the target text element; and 
 in accordance with a determination that the synonymous text element for the target text element is not in the set of reference text elements, obtaining, by the one or more processors, the information describing the target text element. 
 
     
     
       3. The computer-implemented method according to  claim 2 , further comprising:
 in accordance with a determination that the synonymous text element for the target text element is in the set of reference text elements, determining, by the one or more processors, a vectorized representation for the synonymous text element based on a further predetermined association; and 
 determining, by the one or more processors, the target vectorized representation for the target text element based on the vectorized representation for the synonymous text element. 
 
     
     
       4. The computer-implemented method according to  claim 1 , wherein obtaining the information describing the target text element comprises:
 searching, by the one or more processors, for the target text element from a dictionary; and 
 in response to the target text element being found in the dictionary, obtaining, by the one or more processors, the information describing the target text element from the dictionary. 
 
     
     
       5. The computer-implemented method according to  claim 1 , wherein the first machine learning model is a neural network model. 
     
     
       6. The computer-implemented method according to  claim 1 , wherein the second machine learning model is a recurrent neural network model. 
     
     
       7. The computer-implemented method according to  claim 1 , wherein determining the set of vectorized representations for the set of descriptive text elements comprises:
 determining, by the one or more processors, whether the set of descriptive text elements are in the set of reference text elements; 
 in accordance with a determination that a first descriptive text element of the set of descriptive text elements is in the set of reference text elements, determining, by the one or more processors, a first vectorized representation for the first descriptive text element based on the further predetermined association; and 
 in accordance with a determination that a second descriptive text element of the set of descriptive text elements is not in the set of reference text elements, determining, by the one or more processors, a second vectorized representation for the second descriptive text element based on the further predetermined association and a synonymous text element for the second descriptive text element. 
 
     
     
       8. The computer-implemented method according to  claim 1 , wherein the machine learning machine is trained based on a sample vectorized representation for a sample text element and a set of sample vectorized representations for a set of sample descriptive text elements, the set of sample descriptive text elements being comprised in information describing the sample text element. 
     
     
       9. A computer system comprising:
 one or more computer processors; 
 one or more computer readable storage media; 
 program instructions stored on the one or more computer readable storage media for execution by at least one of the one or more computer processors, the program instructions comprising:
 in accordance with a determination that a target text element is not in a set of reference text elements, program instructions to obtain information describing the target text element, the information comprising a set of descriptive text elements and wherein the set of reference text elements are mapped to a set of vectorized presentations based on a further predetermined association and wherein the further predetermined association is represented by a first machine learning model trained based on a corpus comprising the set of reference text elements; 
 program instructions to determine a set of vectorized representations for the set of descriptive text elements further comprising:
 program instructions to determine the set of vectorized representations for the set of descriptive text elements based on the further predetermined association, respectively; and 
 
 program instructions to determine a target vectorized representation for the target text element based on the set of vectorized representations using a second machine learning model, the second machine learning model being trained to represent a predetermined association between the set of vectorized representations for the set of descriptive text elements describing the target text element and the target vectorized representation. 
 
 
     
     
       10. The computer system according to  claim 9 , wherein obtaining the information describing the target text element comprises:
 in accordance with a determination that the target text element is not in the set of reference text elements, program instructions to determine a synonymous text element for the target text element; and 
 in accordance with a determination that the synonymous text element for the target text element is not in the set of reference text elements, program instructions to obtain the information describing the target text element. 
 
     
     
       11. The computer system according to  claim 10 , wherein the acts further comprise:
 in accordance with a determination that the synonymous text element for the target text element is in the set of reference text elements, program instructions to determine a vectorized representation for the synonymous text element based on a further predetermined association; and 
 program instructions to determine the target vectorized representation for the target text element based on the vectorized representation for the synonymous text element. 
 
     
     
       12. The computer system according to  claim 9 , wherein obtaining the information describing the target text element comprises:
 program instructions to search for the target text element from a dictionary; and 
 in response to the target text element being found in the dictionary, program instructions to obtain the information describing the target text element from the dictionary. 
 
     
     
       13. The computer system according to  claim 9 , wherein the first machine learning model is a neural network model. 
     
     
       14. The computer system according to  claim 9 , wherein the further second machine learning model is a recurrent neural network model. 
     
     
       15. The computer system according to  claim 9 , wherein the set of vectorized representations for the set of descriptive text elements comprises:
 program instructions to determine whether the set of descriptive text elements are in the set of reference text elements; 
 in accordance with a determination that a first descriptive text element of the set of descriptive text elements is in the set of reference text elements, program instructions to determine a first vectorized representation for the first descriptive text element based on the further predetermined association; and 
 in accordance with a determination that a second descriptive text element of the set of descriptive text elements is not in the set of reference text elements, program instructions to determine a second vectorized representation for the second descriptive text element based on the further predetermined association and a synonymous text element for the second descriptive text element. 
 
     
     
       16. The computer system according to  claim 9 , wherein the machine learning machine is trained based on a sample vectorized representation for a sample text element and a set of sample vectorized representations for a set of sample descriptive text elements, the set of sample descriptive text elements being comprised in information describing the sample text element. 
     
     
       17. A computer program product comprising:
 one or more computer readable storage media and program instructions stored on the one or more computer readable storage media, the program instructions comprising:
 in accordance with a determination that a target text element is not in a set of reference text elements, program instructions to obtain information describing the target text element, the information comprising a set of descriptive text elements and wherein the set of reference text elements are mapped to a set of vectorized presentations based on a further predetermined association and wherein the further predetermined association is represented by a first machine learning model trained based on a corpus comprising the set of reference text elements; 
 program instructions to determine a set of vectorized representations for the set of descriptive text elements further comprising:
 program instructions to determine the set of vectorized representations for the set of descriptive text elements based on the further predetermined association, respectively; and 
 
 program instructions to determine a target vectorized representation for the target text element based on the set of vectorized representations using a second machine learning model, the second machine learning model being trained to represent a predetermined association between the set of vectorized representations for the set of descriptive text elements describing the target text element and the target vectorized representation. 
 
 
     
     
       18. The computer program product according to  claim 17 , wherein program instructions to obtain the information describing the target text element comprises:
 program instructions to search for the target text element from a dictionary; and 
 in response to the target text element being found in the dictionary, program instructions to obtain the information describing the target text element from the dictionary.